Closed colleenXu closed 2 years ago
Example of logs for semmeddb with timeouts? rate limits? happening with the following query:
Note to @andrewsu @newgene that this issue affects demo queries and may be an issue...
It's worth noting that I've since been unable to replicate my results of getting 503 service unavailable
from BioThings, using a test query that had previously caused such a return when being sent repeatedly in quick succession.
This isn't the case anymore as best as I can tell, so I think my initial claim of rate limiting is now dubious at best. This leaves the timeout issue, which may be on the BioThings end, as @marcodarko has noted relatively simple queries going through normally using GET but taking a long time/not completing through POST.
Closing since this issue hasn't been easily reproducible (isn't a big deal right now) and is somewhat addressed by commits related to batch-querying like this: biothings/bte_trapi_query_graph_handler@21a9c15 and biothings/bte_trapi_query_graph_handler@2313aa5
For sub-queries with many IDs (the later hops of a Predict-style).....it seems like some of the sub-queries BTE sends to biothings and TRAPI APIs fail.
@tokebe has said that:
We'd like to explore this in more detail to see when / why it occurs and if this needs to be addressed and how...